Spaces:
Build error
Build error
Update app.py
Browse files
app.py
CHANGED
|
@@ -2,6 +2,115 @@ import gradio as gr
|
|
| 2 |
import requests
|
| 3 |
import bs4
|
| 4 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5 |
def link_find(url):
|
| 6 |
out = []
|
| 7 |
source = requests.get(url)
|
|
|
|
| 2 |
import requests
|
| 3 |
import bs4
|
| 4 |
|
| 5 |
+
def sort_doc(text,steps_in=0,control=None):
|
| 6 |
+
text=str(text)
|
| 7 |
+
|
| 8 |
+
########################################
|
| 9 |
+
sen_list=get_sen_list(text)
|
| 10 |
+
key_cnt=len(sen_list)
|
| 11 |
+
|
| 12 |
+
sen_obj_box=[]
|
| 13 |
+
for ii,ee in enumerate(sen_list):
|
| 14 |
+
sen_obj=proc_sen(sen_list,ii)
|
| 15 |
+
sen_obj_box.append(sen_obj)
|
| 16 |
+
|
| 17 |
+
sen_list=sen_obj_box
|
| 18 |
+
######################################
|
| 19 |
+
key_cnt=len(sen_obj_box)
|
| 20 |
+
print(key_cnt)
|
| 21 |
+
#noun_cnt=len(noun_box)
|
| 22 |
+
#print(noun_cnt)
|
| 23 |
+
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
if not steps_in:
|
| 27 |
+
|
| 28 |
+
control_char=list(control_json['control'])
|
| 29 |
+
char_len=len(control_char)
|
| 30 |
+
n_cnt=0
|
| 31 |
+
nx=key_cnt
|
| 32 |
+
while True:
|
| 33 |
+
if nx >= 1:
|
| 34 |
+
n_cnt+=1
|
| 35 |
+
nx = nx/char_len
|
| 36 |
+
else:
|
| 37 |
+
print("#######")
|
| 38 |
+
print(n_cnt)
|
| 39 |
+
print(nx)
|
| 40 |
+
print("#######")
|
| 41 |
+
steps=n_cnt
|
| 42 |
+
break
|
| 43 |
+
if steps_in:
|
| 44 |
+
steps=steps_in
|
| 45 |
+
|
| 46 |
+
if control:
|
| 47 |
+
control_len=control_json['leng']-steps
|
| 48 |
+
control_char_val=list(control_json['control'][:control_len])
|
| 49 |
+
control_val=list(control_json['control'][control_len:])
|
| 50 |
+
val_len=len(control_val)
|
| 51 |
+
|
| 52 |
+
|
| 53 |
+
|
| 54 |
+
json_out={}
|
| 55 |
+
noun_list={}
|
| 56 |
+
step_list=[]
|
| 57 |
+
|
| 58 |
+
big_cnt=0
|
| 59 |
+
cnt=0
|
| 60 |
+
go=True
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
step_cont_box=[]
|
| 64 |
+
for ii in range(steps):
|
| 65 |
+
print(ii)
|
| 66 |
+
step_cont_box.append(0)
|
| 67 |
+
#print (step_cont_box)
|
| 68 |
+
mod=0
|
| 69 |
+
pos=len(step_cont_box)-1
|
| 70 |
+
|
| 71 |
+
|
| 72 |
+
|
| 73 |
+
if go:
|
| 74 |
+
for i, ea in enumerate(sen_list):
|
| 75 |
+
if go:
|
| 76 |
+
if cnt > char_len-1:
|
| 77 |
+
#print(step_cont_box)
|
| 78 |
+
go1=True
|
| 79 |
+
for ii,ev in enumerate(step_cont_box):
|
| 80 |
+
if go:
|
| 81 |
+
if ev >= char_len-1:
|
| 82 |
+
step_cont_box[ii]=0
|
| 83 |
+
if go1==True:
|
| 84 |
+
step_cont_box[ii-1]=step_cont_box[ii-1]+1
|
| 85 |
+
go1=False
|
| 86 |
+
cnt=1
|
| 87 |
+
else:
|
| 88 |
+
step_cont_box[pos]=cnt
|
| 89 |
+
cnt+=1
|
| 90 |
+
print(step_cont_box)
|
| 91 |
+
out_js=""
|
| 92 |
+
for iii,j in enumerate(step_cont_box):
|
| 93 |
+
print(j)
|
| 94 |
+
out_js = out_js+control_char[j]
|
| 95 |
+
sen_obj=sen_obj_box[i]
|
| 96 |
+
#sen_obj=proc_sen(sen_list,i)
|
| 97 |
+
|
| 98 |
+
#json_out[out_js]={'nouns':ea}
|
| 99 |
+
json_out[out_js]=sen_obj
|
| 100 |
+
print ("#################")
|
| 101 |
+
print (out_js)
|
| 102 |
+
print (sen_obj)
|
| 103 |
+
print ("#################")
|
| 104 |
+
|
| 105 |
+
big_cnt+=1
|
| 106 |
+
if big_cnt==key_cnt:
|
| 107 |
+
print("DONE")
|
| 108 |
+
go=False
|
| 109 |
+
noun_list=proc_nouns(json_out)
|
| 110 |
+
return json_out, noun_list
|
| 111 |
+
|
| 112 |
+
|
| 113 |
+
|
| 114 |
def link_find(url):
|
| 115 |
out = []
|
| 116 |
source = requests.get(url)
|